Tata Consultancy Services announced this week it's building a workforce of 5,900 to 8,900 "forward-deployed engineers" — roughly 1% to 1.5% of its headcount — to embed with clients and accelerate AI adoption. CEO K Krithivasan framed it as offense, not defense: AI creates new business for an outsourcing giant, it doesn't just automate away the old business. Reasonable bet, and TCS is also shopping for AI, data security, and cybersecurity acquisitions to go with it, after mostly sitting out M&A for years.
Here's the sentence that actually matters, buried in the coverage: Krithivasan did not say whether the company would hire externally or retrain existing staff.
"What you need is a deep knowledge of the customer environment to make it work. That is where we differentiate ourselves. This has nothing to do with cost arbitrage. It's essentially because of the talent pool that we have built."
— K Krithivasan, CEO, Tata Consultancy Services, via Reuters interview, republished in Yahoo Finance
That's not a company being cagey. That's a company that doesn't have the answer yet, because the role doesn't have a template. "Forward-deployed engineer" is barely eighteen months old as a job title with any real currency outside Palantir — Palantir ran the playbook first, OpenAI, Anthropic, and Microsoft have all built out versions of it since, and now the largest IT services companies on earth are chasing the same title with no standardized job description or established comp band, and no settled answer on what background actually predicts success in the seat. TCS is committing to fill up to 8,900 of these roles before anyone — including TCS — has solved the sourcing question.
That gap is not a TCS problem. It's the exact problem sitting on every CTO's desk who's tried to write a forward-deployed engineer req in the last six months.
"FDEs work in a ton of ambiguity, and often what the customer describes in scoping doesn't match the data/system reality on the ground."
— Colin Jarvis, Head of Forward Deployed Engineering, OpenAI, in The Pragmatic Engineer
The title sounds like a solutions architect. It behaves like a consultant embedded on-site with a client's messiest workflows, empowered to write and ship code against their actual systems, on a clock, with the client watching. That's a different animal than the SE or implementation engineer role your comp bands were built for, and treating it like a relabeled version of either one is how you end up six months into a hire who can demo brilliantly in a conference room and freezes the first time a client's legacy stack doesn't cooperate.
Build versus buy versus retrain isn't a philosophical question here — it's the actual decision, and most companies are answering it by default instead of on purpose.
Retrain your best client-facing engineers and you get people who already understand your delivery model, but you're betting they can pick up the technical range fast enough — the range Palantir, OpenAI, and Anthropic have spent years turning into a repeatable playbook — and you're pulling them off the work they're already good at. Hire externally and you get people who've done the job before, but the pool is thin, everyone with real forward-deployed reps is getting bid up by that same handful of firms, and you inherit whoever's playbook they learned on, which may not fit how you sell or deliver.
Most organizations aren't choosing. They're posting a req with a copy-pasted title, running it through the standard technical screen built for a different role, and hoping the pipeline sorts itself out. It doesn't. It produces a hire who looks right on paper and can't survive week three in front of a client.
If TCS — with a recruiting engine that processes hundreds of thousands of hires a year — hasn't locked down build-versus-buy for this role yet, don't assume your internal recruiting function has quietly solved it either. The honest move is admitting the role is new enough that your existing hiring process wasn't built for it, then designing the screen around what the job actually requires: someone who can write production code, read an unfamiliar codebase fast, and hold their own in a room with a skeptical client, all at once. That's a different interview than the one sitting in your ATS template right now. Krithivasan hasn't solved this. Neither have you.
VC5 Consulting helps technology companies define and staff roles that don't have an established playbook yet — forward-deployed engineers, AI implementation leads, and the hybrid technical-client-facing seats that keep showing up faster than job descriptions can keep pace. If you're trying to fill a role your current hiring process wasn't built for, let's talk.